Contact displacement sensor data processing method and system based on big data

By performing component decomposition and merging of Lisaru graphics of contact displacement sensor, combining perfect circle fitting and noise intensity calculation, the problem of inaccurate measurement of grating scales is solved, and data accuracy and processing efficiency are improved.

CN118706005BActive Publication Date: 2024-11-05GUANGZHOU MATE-GRIT INTELLIGENT TECH CO LTD
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202411187035.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-28
Publication Date
2024-11-05
Estimated Expiration
2044-08-28

AI Technical Summary

Technical Problem

The existing contact displacement sensors are inaccurate in measurement due to loose grating scale or contamination during long working hours. The EMD algorithm selects the IMF component when denoising data, resulting in data loss or noise denoising effect, which affects the measurement data. accuracy.

Method used

By obtaining historical measurement results, the Lisa Ru graph is generated, the components are decomposed and combined to obtain the preferred coefficients for denoising, the noise intensity is calculated by combining the perfect circle fitting and interleaving ratio, and the sensor accuracy is adjusted using an abnormality monitoring algorithm.

Benefits of technology

It improves the accuracy and reliability of contact sensor data measurement, enhances abnormal detection capabilities, reduces false alarm rates, and improves data processing efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118706005B_ABST
    Figure CN118706005B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of data processing, and more specifically, to a data processing method and system for a contact displacement sensor based on big data, the method comprising: obtaining historical measurement results to generate a Lissajous figure; decomposing the Lissajous figure to obtain a plurality of components, selecting the components for merging, obtaining a preferred coefficient corresponding to each component of each measurement data, using the preferred coefficient as a weight for denoising, and obtaining a denoised Lissajous figure; obtaining a perfect circular curve of the denoised Lissajous figure, calculating the intersection-and-union ratio of the perfect circular curve and the denoised Lissajous figure, and calculating an abnormal coefficient of each measurement data according to the preferred coefficient and the intersection-and-union ratio; in response to the abnormal coefficient being less than or equal to an abnormal threshold, performing an abnormality detection on the denoised Lissajous figure, thereby adjusting the accuracy of the contact displacement sensor. The present invention corrects the intersection-and-union ratio by using the preferred coefficient, timely discovers abnormal situations, and improves the credibility of the data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of data processing, and more specifically, to a contact displacement sensor data processing method and system based on big data. Background Art

[0002] Contact displacement sensors are essential components for high-end CNC systems because they can obtain accurate displacement data through contact measurement. Existing contact displacement sensors use grating ruler measurement systems for measurement, but after long-term operation, multiple factors may occur, such as loose grating rulers, grating ruler contamination, etc., resulting in inaccurate data measurement of contact displacement sensors.

[0003] The existing Chinese patent application document with publication number CN104949621A discloses a method for locating the boundaries of a grating scale stripe, including: after grayscale transformation of the grating scale stripe image, calculating the vertical projection histogram of the obtained grayscale image; using the empirical mode decomposition (EMD) algorithm to decompose the vertical projection histogram and calculate its trend line; obtaining all pixel positions when the vertical projection histogram is equal to its trend line and establishing an index number set, combining the index number set with the initial pixel position and the end pixel position to generate an index array; calculating the average value of any two adjacent elements in the index array and establishing an average value array; taking two adjacent values ​​in the average value array as an interval, calculating the pixel position corresponding to the maximum value of the second-order derivative of the vertical projection histogram in each interval, and using the obtained pixel position as the edge position of the grating scale stripe image.

[0004] In the above scheme, the vertical projection histogram is only decomposed into several intrinsic mode functions (IMFs) by the empirical mode decomposition (EMD) algorithm. However, the collected data is often noisy, which has a certain interference on the change analysis of the Lissajous figure. Therefore, it is necessary to denoise the Lissajous figure corresponding to the measured data. The existing EMD algorithm can denoise the data, but when the IMF component is selected, if the IMF component is not selected well, it often causes a large data loss or poor noise denoising effect, which can easily lead to inaccurate measurement data, so that subsequent detection will cause errors. Summary of the invention

[0005] In order to solve the problem that when the EMD algorithm performs data denoising, when the IMF component is not well selected, it often causes a large data loss or a poor noise denoising effect, which easily leads to inaccurate measurement data and even causes errors in subsequent detections, the present invention provides solutions in the following multiple aspects.

[0006] In a first aspect, a data processing method for a contact displacement sensor based on big data includes: obtaining historical measurement results to generate a Lissajous figure; decomposing the Lissajous figure to obtain a plurality of components, selecting the components for merging, obtaining a preferred coefficient corresponding to each component of each measurement data, and using the preferred coefficient as a weight for denoising to obtain a denoised Lissajous figure; performing perfect circle fitting on the denoised Lissajous figure to obtain a fitted perfect circle curve, obtaining an intersection-and-union ratio of the perfect circle curve and the denoised Lissajous figure, calculating the noise intensity of each measurement data according to the preferred coefficient and the intersection-and-union ratio, and using the noise intensity as an abnormality coefficient; in response to the abnormality coefficient being less than or equal to an abnormality threshold, using an abnormality monitoring algorithm to perform abnormality detection on the denoised Lissajous figure, thereby adjusting the accuracy of the contact displacement sensor; wherein the abnormality coefficient is: , where Indicates The measurement data anomaly coefficient, Indicates The mean value of the maximum optimal coefficient corresponding to the preset number of measurement data before the measurement data, represents the hyperparameter, Indicates The intersection and union ratio of the measured data is Indicates An exponential function with base .

[0007] The effect is: by decomposing the Lissajous figures into components for denoising, combined with the optimal coefficients, it helps to extract key information from the signal, improves the accuracy and reliability of contact sensor data measurement, enhances the anomaly detection capability, reduces the false alarm rate, and thus improves data processing efficiency.

[0008] Preferably, the Lissajous figure is decomposed to obtain several components, including:

[0009] The Lissajous figure includes displacement data in two directions, wherein the displacement data in the two directions are an X-axis and a Y-axis, and are displayed in a coordinate system in the form of a sine function or a cosine function;

[0010] The Lissajous figure is decomposed by using an EMD algorithm, wherein the EMD algorithm decomposes the signal into a plurality of IMF components, each IMF representing a frequency component of the signal.

[0011] The effect is that the noise components in the signal are separated from the useful signal components through EMD decomposition, and the IMF components obtained by decomposition can be used for anomaly detection. By analyzing the characteristics of specific components, anomalies or mutations in the signal can be identified.

[0012] Preferably, the components are selected and combined to obtain the optimal coefficient of each component corresponding to each measurement data, including:

[0013] The method for selecting the components for merging is: superposition from low frequency to high frequency and from high frequency to low frequency, calculating the correlation between each measured data after superposition and the original signal, and obtaining the optimal coefficient of each measured data corresponding to each component.

[0014] The effect is that by gradually superimposing the components, the reconstruction process of the signal can be optimized to ensure that the reconstructed signal is as close to the original signal as possible. The low-frequency components usually contain the main features of the signal, while the high-frequency components contain the details. The overall characteristics and details of the signal can be balanced by the merging method. By analyzing the signal from different directions (from low frequency to high frequency or from high frequency to low frequency), the adaptability of the signal analysis is provided.

[0015] Preferably, the optimization coefficient satisfies the following relationship:

[0016] ,

[0017] In the formula, Indicates The measurement data The preferred coefficient of each component, For the The measurement data The accumulated components from low frequency to high frequency corresponding to the first component The measurement data The similarity between the accumulated components from low frequency to high frequency corresponding to the components, Indicates The measurement data The accumulated components from high frequency to low frequency corresponding to the first component The measurement data The similarity between the accumulated components from high frequency to low frequency corresponding to the components.

[0018] Preferably, the preferred coefficient is used as a weight for denoising to obtain a denoised Lissajous figure, including:

[0019] Obtain the lower frequency components of the components corresponding to the maximum preferred coefficient, accumulate the lower frequency components, use the product of the maximum preferred coefficient and the corresponding component value as the weight of the accumulated sum of the lower frequency components, reconstruct the signal after superimposing the components, and then iterate to obtain the denoised Lissajous figure.

[0020] The effect is that by iteratively superimposing components, the reconstructed signal is closer to the actual situation of the original signal, and the weighted superposition method can process the signal more efficiently and reduce computational complexity and time.

[0021] Preferably, the denoised Lissajous figure is subjected to perfect circle fitting to obtain a fitted perfect circle curve, including:

[0022] The coordinates of each data point in the Lissajous figure are obtained, a design vector is constructed according to the coordinates of each data point, all the design vectors are arranged in columns, a design matrix is ​​generated, a least squares equation is constructed, and a parameter vector is obtained. The coordinates of the center of the circle and the square of the radius are extracted based on the parameter vector, and a fitting circle is drawn according to the parameters to obtain a perfect circular curve.

[0023] Preferably, obtaining the intersection-over-union ratio of the perfect circular curve and the denoised Lissajous figure includes:

[0024] Fill the inner areas of the perfect circular curve and the denoised Lissajous figure to obtain the areas of the two, calculate the area of ​​the overlapping area between the two figures after filling, calculate the total area after the two are combined, obtain the ratio of the intersection area to the union area, and obtain the intersection-union ratio of the perfect circular curve and the denoised Lissajous figure; wherein the total area is the area of ​​the non-overlapping part and the intersection part, and the denoised Lissajous figure is a smooth figure after removing the irregular part.

[0025] The effect is that by comparing the intersection-and-union ratio, the difference in shape between the perfect circular curve and the denoised Lissajous figure can be identified, so as to understand the possible deviations in the fitting process. When the intersection-and-union ratio is low, it indicates that the noise has not been completely eliminated during the denoising process or there are other problems. The denoising algorithm needs to be further optimized to improve the credibility of the data.

[0026] In a second aspect, a contact displacement sensor data processing system based on big data includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned contact displacement sensor data processing method based on big data is implemented.

[0027] The present invention has the following effects:

[0028] 1. The present invention selects components through the changes of high- and low-frequency accumulated components between continuous measurement data, and performs weighted superposition according to the preferred coefficients based on the components with the largest preferred coefficients, thereby ensuring the robustness of the denoising effect when modes overlap.

[0029] 2. The present invention performs perfect circle fitting on the denoised Lissajous figure, obtains the intersection-and-union ratio between the denoised Lissajous figure and the standard circle, and judges the difference between the Lissajous figure and the standard circle. In order to reduce the misjudgment caused by comparing the intersection-and-union ratio, the abnormal coefficient is obtained based on the optimal coefficient and the intersection-and-union ratio. The abnormal coefficient is used as a judgment standard to timely discover abnormal situations, thereby calibrating the contact sensor and improving the credibility of the data. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0031] Figure 1 It is a method flow chart of steps S1 to S4 in the contact displacement sensor data processing method based on big data in an embodiment of the present invention.

[0032] Figure 2 It is a method flow chart of steps S20 to S22 in the contact displacement sensor data processing method based on big data in an embodiment of the present invention.

[0033] Figure 3 It is a structural block diagram of a contact displacement sensor data processing system based on big data according to an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0035] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0036] Reference Figure 1 The data processing method of the contact displacement sensor based on big data includes steps S1 to S4, which are as follows:

[0037] It should be noted that the contact displacement sensor in this solution is a grating ruler, which is a high-precision measurement tool that uses the movement of grating stripes to measure displacement. It is widely used in the field of precision measurement, especially in high-end equipment such as CNC machine tools, robotics, and precision measuring instruments.

[0038] The working principle of the grating ruler is as follows: when light passes through the grating, the transparent and opaque parts on the grating interfere with the light source and the observer, generating moiré fringes. Moiré fringes are the result of the interaction between two sets of periodic structures. Among them, the photosensitive element (such as a photodiode or a phototransistor) is sensitive to the brightness changes of the moiré fringes; when the bright line of the moiré fringes passes through the sensor, the sensor receives more light and generates a higher output signal; when the dark line passes, the received light decreases and the output signal is lower. As the grating ruler moves, the position and shape of the moiré fringes change, causing the light intensity received by the photosensitive element to change accordingly. This change can be converted into an electrical signal that changes with time, thereby realizing the measurement of the grating ruler, in which the measured electrical signal is a sinusoidal signal. Through signal processing techniques such as filtering, denoising, signal amplification, etc., the electrical signal is converted into displacement data. These data can be used to control the movement of the device or measure the displacement of an object.

[0039] S1: Obtain historical measurement results to generate Lissajous figures.

[0040] Further explanation: According to the movement of the grating ruler, a sinusoidal signal is formed, and the multi-phase sinusoidal signals are superimposed to obtain a Lissajous figure. The Lissajous figure includes displacement data in two directions, wherein the displacement data in the two directions are the X-axis and the Y-axis, and are displayed in the coordinate system in the form of a sine function or a cosine function.

[0041] S2: Decompose the Lissajous figure to obtain several components, select the components to merge, obtain the optimal coefficient of each component corresponding to each measurement data, use the optimal coefficient as a weight for denoising, and obtain the denoised Lissajous figure.

[0042] It should be explained that, at present, the Pearson coefficient between each component and the original data is calculated, and whether to retain the component value is determined according to the value of the Pearson coefficient. However, when the EMD algorithm is decomposed, the component is unstable, which may cause the low-frequency component to have a smaller Pearson correlation coefficient with the original signal, wherein a smaller Pearson correlation coefficient indicates a lower correlation. In order to better merge the components, in this embodiment, a method of superimposing the components from low frequency to high frequency is selected, so that the correlation between the superimposed data and the original signal gradually increases. However, since the original signal itself contains noise interference, this solution chooses not to use the correlation with the original signal as a basis for judgment. Moreover, since the low frequency often contains a lot of information in the data, and in continuous measurement, although the measurement length may be different, the accuracy change in the measurement process will not be too large, and the Lissajous figures in continuous measurement should be similar, and then the correlation between the continuously accumulated components is selected as a basis for judgment to improve the accuracy of component merging, as follows:

[0043] Reference Figure 2 , including steps S20 to S22:

[0044] S20: Decomposing the Lissajous figure using an EMD algorithm, wherein the EMD algorithm decomposes the signal into a plurality of IMF components, each IMF representing a frequency component of the signal.

[0045] Further explanation, in this embodiment, obtaining the Lissajous figure data corresponding to the measured data ,in is a time series data, and the EMD algorithm is used to The data is linearly decomposed to obtain multiple IMF components, where the number of decompositions of the IMF components is set to 10 in this scheme and can be adjusted by the implementer according to the specific implementation scenario.

[0046] S21: The method of selecting components for merging is: superposition from low frequency to high frequency and from high frequency to low frequency, calculating the correlation between each measurement data after superposition and the original signal, and obtaining the optimal coefficient of each measurement data corresponding to each component.

[0047] Specifically, the coefficients preferably satisfy the following relationship:

[0048] ,

[0049] In the formula, Indicates The measurement data The preferred coefficient of each component, For the The measurement data The accumulated components from low frequency to high frequency corresponding to the first component The measurement data The similarity between the accumulated components from low frequency to high frequency corresponding to the components, Indicates The measurement data The accumulated components from high frequency to low frequency corresponding to the first component The measurement data The similarity between the accumulated components from high frequency to low frequency corresponding to the components.

[0050] Further explanation, the accumulated components from low frequency to high frequency are The accumulated components of all lower frequency components and residual components after the first component. Similarly, the accumulation from high frequency to low frequency is the first The cumulative component of all higher frequency components and residual components after the component, wherein the method of obtaining the similarity value between two components in this scheme is: using the Pearson correlation coefficient to obtain, which is a well-known technology in the field and will not be described in detail; wherein the larger the Pearson correlation coefficient value, the stronger the correlation, and the negative number indicates a negative correlation, and then as the components are accumulated, the higher the correlation coefficient will be;

[0051] Since noise is random, the lower its correlation, the higher its preference, and the more it should be removed as noise. A negative correlation mapping is performed, where a Pearson coefficient value of 0 indicates no correlation, a Pearson coefficient value of -1 indicates a negative correlation, and a Pearson coefficient value of 1 indicates a positive correlation, so Complete the pair Negative correlation mapping.

[0052] S22: Obtain the lower frequency components of the components corresponding to the maximum preferred coefficient, accumulate the lower frequency components, use the product of the maximum preferred coefficient and the corresponding component value as the weight of the accumulated sum of the lower frequency components, reconstruct the signal after superimposing the components, and then iterate to obtain the denoised Lissajous figure.

[0053] Further explanation, and then obtain the The optimal coefficients corresponding to each component in the measured data are obtained to obtain the component corresponding to the maximum optimal coefficient , in response to the component corresponding to the maximum preferred coefficient If there is no noise in The measurement data and The measured data are consistent, and then the optimal coefficient approaches 1, and the component corresponding to the maximum optimal coefficient The more noise content in the , the lower the preferred coefficient, where the component corresponding to the maximum preferred coefficient Higher-frequency components are regarded as noise components and need to be discarded.

[0054] In addition to the above solution, in another embodiment, the Lissajous figure can also be decomposed by wavelet transform, and the steps are as follows:

[0055] Obtain the periodic signal in the two sine waves of the Lissajous figure, perform multi-level wavelet decomposition of the Lissajous figure through Morlet wavelet, analyze the coefficients after wavelet transform to identify the main components and noise of the signal, where the high-frequency part contains more noise, while the low-frequency part contains the main information of the signal;

[0056] A threshold is determined according to the median absolute deviation of the wavelet coefficients, and the wavelet coefficients are processed. In response, the wavelet coefficients above the threshold are retained, and conversely, the wavelet coefficients below the threshold can be set to zero or adjusted to remove noise;

[0057] The processed wavelet coefficients are used to perform inverse wavelet transform to reconstruct the signal and obtain the denoised Lissajous figure.

[0058] Among them, the median absolute deviation of the wavelet coefficients determines the threshold as follows: ,in, represents the threshold value, represents the median absolute deviation, Indicates the signal length; MAD (Median Absolute Deviation) is a well-known technology in this field and will not be described in detail.

[0059] S3: Perform perfect circle fitting on the denoised Lissajous figure to obtain a fitted perfect circle curve, obtain the intersection-and-union ratio of the perfect circle curve and the denoised Lissajous figure, calculate the noise intensity of each measurement data according to the preferred coefficient and the intersection-and-union ratio, and use the noise intensity as the abnormal coefficient.

[0060] Further explanation, since the ideal standard Lissajous figure is a standard perfect circle, when an abnormality occurs, it will become an ellipse, and then the denoised Lissajous figure is fitted with a perfect circle. In this embodiment, the least squares method is used to fit the perfect circle, and the fitted perfect circle curve and the denoised Lissajous figure are obtained, and the intersection and union ratio of the perfect circle curve and the denoised Lissajous figure is obtained, where the smaller the intersection and union ratio, the greater the error. However, with the use of contact displacement sensors, the noise may become larger or increase due to local conditions, resulting in false detection based on the intersection and union ratio, so it is necessary to optimize according to the preferred coefficient. The specific steps are as follows:

[0061] Get the coordinates of each data point in the Lissajous figure, construct a design vector based on the coordinates of each data point, arrange all design vectors in columns, generate a design matrix, construct a least squares equation, get a parameter vector, extract the center coordinates and the square of the radius based on the parameter vector, draw a fitting circle based on the parameters, and get a perfect circular curve.

[0062] Fill the inner areas of the perfect circular curve and the denoised Lissajous figure to obtain the areas of the two, calculate the area of ​​the overlapping area between the two figures after filling, calculate the total area after the two figures are combined, obtain the ratio of the intersection area to the union area, and obtain the intersection-union ratio of the perfect circular curve and the denoised Lissajous figure;

[0063] In this embodiment, the intersection-over-union ratio is a technology well known to those skilled in the art and will not be described in detail, wherein the total area is the area of ​​the non-overlapping part and the intersection part, and the denoised Lissajous figure is the smoothed figure after removing the irregular parts.

[0064] Specifically, the anomaly coefficient satisfies the following relationship:

[0065] ,

[0066] In the formula, Indicates The measurement data anomaly coefficient, Indicates The mean value of the maximum optimal coefficient corresponding to the preset number of measurement data before the measurement data, represents the hyperparameter, Indicates The intersection and union ratio of the measured data is Indicates An exponential function with base .

[0067] In this embodiment, Indicates that the first 10 measurement data correspond to , where 10 is a hyperparameter, indicating The noise factor of the measured data is When the value is 1, it means that the noise does not exist, and then the intersection-and-union ratio should be relied upon. .Right now: , when noise exists, the intersection-over-union ratio should be increased to a certain extent, because the lower the intersection-over-union ratio, the more abnormal it is, but it may be the false interference caused by noise, and then a hyperparameter is set in the formula , this solution sets , which can be adjusted by implementers according to specific implementation scenarios.

[0068] S4: In response to the abnormality coefficient being less than or equal to the abnormality threshold, an abnormality monitoring algorithm is used to perform abnormality detection on the denoised Lissajous figure, thereby adjusting the accuracy of the contact displacement sensor.

[0069] To further illustrate, the anomaly coefficient threshold is set to 0.97. When the anomaly coefficient is less than or equal to the anomaly coefficient threshold, anomaly detection is performed. The anomaly detection can be performed by using an existing anomaly monitoring algorithm to monitor the accuracy of the measured data. The anomaly monitoring algorithm is, for example, LDA (Linear Discriminant Analysis), neural network, random forest, etc.

[0070] At this point, the method of the present invention ends here.

[0071] The present invention also provides a contact displacement sensor data processing system based on big data. Figure 3 As shown, the system includes a processor and a memory, the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the contact displacement sensor data processing method based on big data of the first aspect of the present invention is implemented.

[0072] The system also includes other components familiar to those skilled in the art, such as a communication bus and a communication interface. The configuration and functions of these components are known in the art and will not be described in detail here.

[0073] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.

[0074] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0075] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A contact displacement sensor data processing method based on big data, characterized in that: include: Obtain historical measurement results to generate Lissajous figures; Decomposing the Lissajous figure to obtain a plurality of components, selecting the components for merging to obtain a preferred coefficient of each component corresponding to each measurement data, and using the preferred coefficient as a weight for denoising to obtain a denoised Lissajous figure; The method of selecting the components for merging is: superposition from low frequency to high frequency and from high frequency to low frequency, calculating the correlation between each measured data after superposition and the original signal, and obtaining the optimal coefficient of each measured data corresponding to each component; Performing perfect circle fitting on the denoised Lissajous figure to obtain a fitted perfect circle curve, obtaining an intersection-and-union ratio of the perfect circle curve and the denoised Lissajous figure, calculating the noise intensity of each measurement data according to the preferred coefficient and the intersection-and-union ratio, and taking the noise intensity as an abnormality coefficient; In response to the abnormal coefficient being less than or equal to the abnormal threshold, an abnormality detection algorithm is used to perform abnormality detection on the denoised Lissajous figure, thereby adjusting the accuracy of the contact displacement sensor; Among them, the abnormal coefficient is: , In the formula, Indicates The measurement data anomaly coefficient, Indicates The mean value of the maximum optimal coefficient corresponding to the preset number of measurement data before the measurement data, represents the hyperparameter, Indicates The intersection and union ratio of the measured data is Indicates An exponential function with base .

2. The contact displacement sensor data processing method based on big data according to claim 1 is characterized in that: The Lissajous figure is decomposed to obtain several components, including: The Lissajous figure includes displacement data in two directions, wherein the displacement data in the two directions are an X-axis and a Y-axis, and are displayed in a coordinate system in the form of a sine function or a cosine function; The Lissajous figure is decomposed by using an EMD algorithm, wherein the EMD algorithm decomposes the signal into a plurality of IMF components, each IMF representing a frequency component of the signal.

3. The contact displacement sensor data processing method based on big data according to claim 1, characterized in that: The optimization coefficient satisfies the following relationship: , In the formula, Indicates The measurement data The preferred coefficient of each component, For the The measurement data The accumulated components from low frequency to high frequency corresponding to the first component The measurement data The similarity between the accumulated components from low frequency to high frequency corresponding to the components, Indicates The measurement data The accumulated components from high frequency to low frequency corresponding to the first component The measurement data The similarity between the accumulated components from high frequency to low frequency corresponding to the components.

4. The contact displacement sensor data processing method based on big data according to claim 1, characterized in that: The preferred coefficient is used as a weight for denoising to obtain a denoised Lissajous figure, including: Obtain the lower frequency components of the components corresponding to the maximum preferred coefficient, accumulate the lower frequency components, use the product of the maximum preferred coefficient and the corresponding component value as the weight of the accumulated sum of the lower frequency components, reconstruct the signal after superimposing the components, and then iterate to obtain the denoised Lissajous figure.

5. The contact displacement sensor data processing method based on big data according to claim 1, characterized in that: The denoised Lissajous figure is subjected to perfect circle fitting to obtain a fitted perfect circle curve, including: The coordinates of each data point in the Lissajous figure are obtained, a design vector is constructed according to the coordinates of each data point, all the design vectors are arranged in columns, a design matrix is ​​generated, a least squares equation is constructed, and a parameter vector is obtained. The coordinates of the center of the circle and the square of the radius are extracted based on the parameter vector, and a fitting circle is drawn according to the parameters to obtain a perfect circular curve.

6. The contact displacement sensor data processing method based on big data according to claim 1, characterized in that: Obtaining the intersection-and-union ratio of the perfect circular curve and the denoised Lissajous figure, including: Fill the inner areas of the perfect circular curve and the denoised Lissajous figure to obtain the areas of the two, calculate the area of ​​the overlapping area between the two figures after filling, calculate the total area after the two figures are combined, obtain the ratio of the intersection area to the union area, and obtain the intersection-union ratio of the perfect circular curve and the denoised Lissajous figure; The total area is the area of ​​the non-overlapping part and the intersection part, and the denoised Lissajous figure is a smooth figure after removing the irregular parts.

7. A contact displacement sensor data processing system based on big data, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the data processing method of the contact displacement sensor based on big data according to any one of claims 1 to 6 is implemented.

Citation Information

Patent Citations

  • Boundary positioning method of stripes of grating ruler

    CN104949621A

  • Mechanical dithering laser gyroscope light-combined adjusting system

    CN103454925A

  • Local abnormal factor optimization ellipse fitting method for optical fiber vibration sensing phase demodulation

    CN114812636A

  • Noise suppressor

    US20030128851A1

  • Real-time normalization apparatus and method of phase generated carrier demodulation in sinusoidal phase modulation interferometer

    US20190368860A1